Having studied the last year with a lecturer who is in love with researching HRTFs (Head Related Transfer Functions) I have a few personal thoughts on binaural recording and mixing.
In the lab we had our HRTFs measured. The way we did it was stick a microphone facing out of our ears and record the sound of several speakers at varying heights playing a short sine sweep. After processing and analysis you end up with a frequency response that shows how your head responds to sounds.
This showed me that certain notches and peaks appear when sound is heard from different angles of elevation.
It gave me some good ideas to experiment with filtering those freqs when designing sounds to evoke the feeling that a sound was localised at a certain elevation.
We also ran through quite a few tests where the teacher played back sounds to us over headphones that he had convolved with a range of different peoples HRTFs.
The idea being to use a GUI to say whether one sound was higher or lower in elevation to the previous sound.
This and other tests pointed out to me the two main problems I have with HRTFs or binaural recordings. 1) Everyones head and ears are of different sizes so produce frequency notches and peaks at different points, which therefore affect our localisation skills of sound sources. 2) There are lots of aural blind spots with the sounds of binaural recordings, the main one for me being that sound localisation for front and back sources is difficult, some people in tests said a sound was infront while others heard the same sound at the rear.
However I made a radio play utilising a dummy head. It was a creepy scary story and from what we recorded the best sounds came from being up close to the head, I guess where there is a distinct time and level delay produced by the interference of the head.
I know the above doesn't really answer your question, but I would consider doing some research on HRTFs and using that information to what makes binaural recordings work. Maybe I've been overloaded with learning all this shit the last year that I think buying equipment might be the wrong way to go, considering we went to the extent of removing any colouration of the sound that could come from the microphones, speakers and headphones just so we could get the truest binaural response we could achieve and then I still wasn't impressed with the outcomes.
It was great fun recording with a dummy head though!